Time-bounded cooperative recovery with the distributed real-time conversation scheme

Of several schemes proposed to handle the propagation of erroneous information among interacting processes in distributed and parallel computer systems, the distributed real-time conversation (DRC) scheme stands out in its fast forward recovery capability which is essential in safety-critical hard-real-time applications. However, previous formulations of the scheme remained at relatively abstract levels and practical models for their implementation in complex safety-critical real-time applications were not established before. The core approach in the DRC scheme is to make a group of computing stations cooperate in recovery from hardware and software faults that may occur during their interaction. In this paper we present a practical implementation model for the DRC scheme. A simple model of an anti-missile defense system is used to illustrate the main structuring principles of the DRC scheme and major components of the practical implementation model.